Knowledge Management in M-SMEs

Knowledge can be referred as information processed by comprehension and the individual's experiences ¹. It contains information that is available and can be used in making decisions and actions ². Knowledge in two dimensions - tacit and explicit ³. Is suggested that knowledge is tacit if it is tied to the senses, experiences and intuition ⁴. Explicit knowledge is characterized as transmittable in formal language and can be stored in information systems ⁵.

Knowledge management is the process of taking measures to have an effective use of assets to acquire strategic advantage ⁶. The main objective of KM is to ensure vital information (knowledge) is properly identified, selected, organized, shared and published to help a firm to perform its operations and achieve its objectives ⁷.

Regarding most M-SMEs, they manifest as less experience organizations that do not have a defined strategy nor understating of how they take advantage of their knowledge assets, as oppose of the larger ones ⁸. These enterprises can benefit from KM activities about employee development, innovation, customer satisfaction and organizational success, since they tend to place more emphasis than larger firms on the management of tacit knowledge as they usually don't have the resources to invest on advanced technologies ⁹. The M-SME sector needs to acquire knowledge in order to implement successful strategies and improve its capabilities ¹⁰.

Often, M-SMEs must deal with specific challenges which strongly differ from those faced by big companies. However, it is quite common, to adopt techniques which have been tried and tested in big companies ¹¹.

Design of databases as a storage technology

Database design is a complex decision-making process. Each element has many characteristics and performances required to make movement between parameters easier. To do an effective design, it is necessary to develop a tool for collecting, storing and processing large amounts of data. The most convenient tool to perform these tasks is a database ⁽¹¹. They also state that during its design, it is necessary to use the most relevant data on the subject and establish their relations, which will allow to operate various types of data, such as dates, numbers, currency and text.

Extracting data from a wide database requires that complex queries should be built across different business objects, which can only be accomplished by database specialists. Using an operational system for decision support has become even more inefficient with today's increasing volume and complexity of data. For decision support, when real-time ad hoc complex queries are required for interactive analysis, a multidimensional structure that contains historical data, aggregates relevant information from various sources, and is maintained separately from operational databases is needed. This structure is known as a data warehouse. It is defined as the combination of technologies that facilitates and maintains the management of historical data provided from different sources ¹².

The purpose of designing databases, thus, turned out to be how to acquire, classify, store, and analyze huge datasets and extract knowledge from them. Furthermore, the processing and the analysis of massive data has additional challenges, such as how to address this data, how to speed up the processing, and how to maintain the data veracity. To extract and process interest data, it is recommended to use the process known as Knowledge Discovery in Databases (KDD) process by which the data are selected, preprocessed, transformed, mined, and evaluated ¹³.

It is considered that creating a database is primarily preferred for complex domains if design involves a lot of domain experts and a group of analyst-designers ¹⁴, but as a result, a unified irredundant database scheme should be obtained. In addition, the database design is a continuous process: client needs, a detailed description of domain entities, their attributes and how those entities relate to each other.

Database normalization

Database normalization is a key process for designing databases. Normalization is a process that takes an existing database design that can have anomalies, errors o redundancies, and improves it by decomposing relations, based on a known set of dependencies. When the normalization is not applied, database systems are usually inefficient and inaccurate ¹⁵.

Relational database relationship must meet certain requirements, to meet the different paradigms that database design usually implies ¹⁶. At the time of writing this article, there are six kinds of relationship database paradigm: the first normal form (1NF), the second normal form (2NF), the third normal form (3NF), BC normal form (BCNF), the fourth normal form (4NF), the fifth normal form (5NF) and the sixth Normal Form (6NF). To fulfill normal requirements, the 1NF is the minimum paradigm. Basing on the 1NF, further meeting more requirements is the 2NF and a higher-level paradigm. Fortunately, in practical design application, you normally only need to reach the 3NF.

Systematic Layout Planning

Systematic Layout Planning (SLP) is a prominent procedural approach and is widely used in layout design for various small and medium enterprises, and although it's a bit of and older technique, recent works have been expanding it use to other areas of application because the advantages it offers as a layout design tool ¹⁷.

From another perspective, the restructuring of the layout is done to optimize the work process, always ensuring the security of the flow of materials, people and information ¹⁸.

SLP is a procedural layout design approach and is a relatively straight forward method that has been a proven tool in providing layout design guidelines ¹⁹, and although traditionally it has been used for designing facility layouts, the fact that it has

been proven in practice makes it an ideal candidate to use for other applications where designing a layout it's the main desire. The more traditional application of SLP method is used to determine a reasonable production system layout scheme that is based on the analysis of a logistic relationship between blocks and operating unit relationships, then, uses a factor evaluation method to assess the schemes, and finally obtains an optimal scheme for the new layout.

There are three main steps in SLP: relationship diagram, Space relationship diagram and evaluation. The relationship diagram shows the importance of each department/ area with respect to each other. It includes logistics relationships diagram, non-logistics relationships and comprehensive relationships. When the goal is to achieve an effective layout, it needs to begin with an in-depth discussion of work relationships. In this case, the primary focus of the discussions surrounded the issue closeness. Each unit was related to every other unit in the system. In the relationship diagram, these closeness values were placed in the corresponding diamond. The closeness rating is shown in Figure 1 ²⁰.

Fuente: Lin et al., 201320

Figure 1 Closeness rating table for SLP. 

Entity-Relationship Diagram

To develop a conceptual data model, normally it involves analyzing the problem domain and coming up with a database diagram using the Entity-Relationship (ER) model. Once the diagram is designed, it is then converted into a relational schema before the actual database is created and implemented. However, although it may sound easy to map an ER diagram, it is not an easy task to perform, it is not as easy for a novice, as it needs to meet some requirements and it involves complex logical thinking ²¹.

For database design and system analysis purposes, the entity-relationship (ER) model and its accompanying ER diagrams (ERD) are the ones used ²². ERD provides developers with an overall grasp of the data requirements, modeling and database structures of the information system before the implementation phase.

Phase 1: Establish the case studies relevant data

On this phase, after getting an amount of case studies of KM application on SMEs, an analysis will be realized to detect the principal and frequent characteristics presented on those. Those ones are the attributes. The attributes on the database represents the name of each column.

Phase 2: Design the Entity Relationship Diagram (ERD)

To design the ERD, first, the Systematic Layout Planning will be realized. This activity consists of assigning a value of each possible relation on a group of elements, and those elements will be the attributes. After the value assignation, the attributes will be aggrouped according of its most valuable relations, resulting in the definition of the entities, which are the tables name. Once defined, the 3 forms of normalization will be executed to get the appropriate relationships between each entity.

Phase 3: Define the attributes characteristics

Based on the ERD design and the requirements of the project, the Database attributes characteristics will be planned. Each attribute needs to have a type of input, the restrictions and if it is a primary key or not.

Phase 4: Develop the database

Once the ERD is ready and the attributes are well defined, the Database System will be chosen. The system works as the development environment, where the entities, their relationships, the attributes and their characteristics will be developed.

Phase 5: Feed the case studies data on the Database

After the database structure is developed, it will be ready to be feed. Based on each case study, the required information will be extracted and inserted on their corresponding attribute, in a way where specific filters can be executed to get the required information.

Phase 1: Established case studies relevant data

To get the relevant data, a series of case studies needed to be analyzed. They were selected by: the year of publication (last 10 years), indicators that define a SME (number of employees and years in the market), thematic (about KM, innovation or intellectual capital applied on SMEs), and the publishing company (a worldwide known journal). Those articles were read, where focus was on data that would let the reader know the actual state of the company regarding KM, those being: the company KM practices such as tools, strategies, areas involved, lessons learned; and information regarding their key activities and characteristics. Table 1 indexes the characteristics considered to search on the journals database.

Table 1 Data to be extracted from case studies. 

Phase 2: Design the Entity Relationship Diagram (ERD)

The ERD will be design after establishing what will be the entities (tables), the attributes (columns names) and the relationship between entities. To determine the members of each element, the aspects defined on the previous phase will be aggrouped to find its common characteristic, and how they relate to each other. The aggrupation consists of using the relationship diagram and the distribution layout model adapted from the Systematic Layout Planning (SLP). The first one is created by assigning a closeness rating of relation between each characteristic, as shown in Figure 3.

Figure 3 Relationship diagram with the closeness rating assigned in each pair of attributes. 

After getting the relations, a distribution layout can be built, presented on Figure 4. By getting the characteristics with the maximum value together, a group was formed to represent those elements, resulting in 4 groups: KM practices, business, articles and lessons learned. The previous layout is the base to design the database. Since the attributes and entities are defined, a normalization is required to ensure the database won't present issues that could lead to data integrity loss. To normalize, the database must be in 3 different normal forms or "NF" (1NF, 2NF and 3NF).

Figure 4 Distribution layout reflected on the closeness rating, by getting the attributes with the highest values closer and further the ones with the lowest. 

The 1NF will be reached if none of the entities have rows with multiple values, so, the case study entity will be the link between the other 3 ones, by adding an attribute as primary key of "id_case_study" and declaring that this entity has a relationship of one-to-many with each of the rest of them, that means the case study can have many KM practices, business and lessons learned. Regarding 2NF, it is necessary that a database non-key attributes depend only of one primary key and met the requirements of 1NF. In that case, the Business entity gets a primary key called "id_business" and as KM practices as Lessons learned entities, they both have a composite primary key, being "id_case_study" the main one, and the other being "id_business". The 3NF has the requirement that there is not dependency between non-key attributes, which this database meets, because each entity non-key attributes depends only of one primary key. After the normalization and using the layout of Figure 3, the ERD results in the one shown in Figure 5.

Figure 5 ERD designed for the database. Being the entities represented as rectangles, their attributes as ovals and their relations as diamonds. 

Phase 3: Define the attributes characteristics

By getting a well-designed database, it is necessary to set every attribute in a way a database can work with. To make that happen, database systems ask the user for a series of options to select, which are described as:

• Column name: A name is assigned to the attribute, must be without spaces and if necessary, add a hyphen.

• Datatype (DT): The user must select from a variety of different data types as integers, Boolean, date or char. Most of them ask the user for the size of that variable, usually is by characters allowed.

• Is primary key (PK): Determines if the attribute acts as primary key or not.

• Not null (NN): Adds a condition of only accept cells when a non-null value is assigned.

• Unique index (UQ): Conditions each row to have a unique value.

• Is binary column (B): Determines if he char or varchar datatype will be binary or not.

• Is unsigned datatype (UN): Asks if the numeric datatype will allow negative numbers.

• Set "0" as default value (ZF): Assigns the value zero if not specified.

• Auto incremental (AI): For each row added, the corresponding value will be the consecutive.

• Generated column (G): All values on the column will be resulted from a command at creation.

• Default/Expression (D/E): The user enters the value assigned if there is not an assignment or, if the column is generated, the corresponding expression.

Given the previous description, all attributes will be set by assign them the required options. Table 2 present the selected options, those will be the final setting to star developing the database.

Table 2 Attributes and the options selected to define the values contained on each of them. 

Phase 4 Develop the database

Once the database is designed, a database system needs to be chosen. For this project, the selected system is MySQL, with a software called MySQL Workbench, version 6.3 CE. The system was chosen because there is more knowledge of its use in comparison with other systems and provides visual tools that facilitates the development. With the data presented on the previous phases, the database is created, with its model presented on Figure 6, where the entities were assigned, just as their attributes and those attributes that creates a link between two entities, selecting if that relation is one-to-one, one-to-many or many-to-many. Now that the database is set, the data extracted from case studies will be inserted on it.

Phase 5: Feed the case studies data on the Database

Based on the criteria defined on phase 1 and 25 case studies which met the requirements, data were extracted from them and then inserted on the corresponding attributes fields. With the database and its data ready, the researcher can now make queries with the conditions he/she consider fit to receive the required data without getting repeated rows or empty fields. Figure 7 shows an example of one query, which requests the article name, the business activity, the KM tools and the KM area where the tool belongs, under a condition where the KM area value must be "Knowledge Sharing".